In partnership with
Your Documentation Just Became a Growth Channel
AI agents don't browse like humans. They analyze everything, compare competitors, and make recommendations at scale.
The companies winning the AI era are optimizing their docs for agents, not just users. Discover how to make your product easier to find, understand, and recommend.
Today In Ai
1. AI chip selloff wipes $1+ trillion off US stocks. Disappointing earnings from Broadcom, Marvell, and Micron sent chipmakers down double digits last week, ending the S&P 500's nine-week winning streak. Read the breakdown.
2. ChatGPT is about to look very different. OpenAI is reportedly turning ChatGPT into a superapp bundling Codex, AI agents, image generation, and third-party tools in one place. Alongside that, they've rolled out Lockdown Mode for prompt-injection protection, a smarter memory system, and the ability to send emails directly from chat.
3. Google to pay SpaceX $920M per month for AI compute. A new SEC filing reveals Google will pay SpaceX nearly $1 billion per month through 2029 for access to 110,000 Nvidia GPUs. The deal drops right before SpaceX's market debut on June 12. Read more.
From The Frontier
More apps are launching than ever, but total app usage has actually fallen since 2025. A viral Financial Times chart made that case clearly last weekend.
Building is the easy part now. Vibe coding tools can take you from idea to live product in hours. The hard part, distribution, getting real people to discover and stick with your app, hasn't gotten easier at all. It's arguably gotten harder.
The approaches that actually work: building an audience before you launch, budgeting for paid ads, or finding a creative hook that earns organic attention.
The proof is already out there. Derrick Downey Jr. vibe coded DualShot Recorder and hit No. 1 on the App Store within 24 hours. Alex Finn's Creator Buddy found its audience too. And a group of high schoolers built CalAI, an AI calorie tracker that hit $30M in annual revenue before getting acquired. Code is cheap. Attention is not.
Karpathy on real AI use: A viral tutorial on how OpenAI co-founder Andrej Karpathy actually uses AI day to day has racked up 16K bookmarks. Worth a read if you want a practitioner's take, not a marketer's.
Gemini edits through your camera: Gemini Live now lets you point your camera at something and edit it in real time with a prompt. Four million views. People are paying attention.
A Pokedex for real animals: Someone built a Pokemon Go-style app that lets you collect real animals you spot in the wild. The waitlist is open at gotcha.jurre.me.
Claude usage doubled: Anthropic has doubled usage limits for paid Cowork plans through July 5th. If you're on a paid plan, now's the time to push it.
Claude prompts itself: Claude Code's creator says he no longer writes prompts manually. He uses looping to let Claude prompt itself. 1.5M views. Yes, we're all thinking the same thing.
Prompt Station
Your weekly planning prompt. Copy. Paste. Done.
This prompt by NanoBanana2 has been making the rounds for a reason. Drop in your favorite show's filming location and watch it turn into a tiny, photorealistic toy-world. It works with any AI image generator.
COPY + PASTE THIS PROMPT
Hyper-realistic miniature diorama of [TV Location], the building fills 70% of the frame, shot from a close elevated isometric angle, tilt-shift lens blur in the foreground and background creating a toy-like miniature effect, warm golden hour lighting, highly detailed scale model aesthetic, surrounding environment soft and secondary, photorealistic, cinematic quality, 4K, 4:5 aspect ratio.
We killed RAG and sandboxes. Here's what we built instead.
Our documentation assistant had a problem. RAG pipelines only sent the model page fragments, making responses feel half-baked. The fix was obvious: give the agent full doc access. Quality jumped, but startup hit 46 seconds and costs ballooned to $70K+/year.
The insight: the agent doesn't need a real filesystem. It just needs to think it's in one.
We built ChromaFs, a virtual filesystem that translates standard UNIX commands into queries against our Chroma database. Every doc page becomes a file, every section a directory. The agent explores documentation the way a developer explores a codebase.
The results: startup dropped from 46s to ~100ms, marginal compute cost hit $0 per conversation, and output quality matched the full sandbox, with access control built in.
ChromaFs now powers 30,000+ daily conversations for hundreds of thousands of users. No containers, no cold starts, no invoice surprises.
Mintlify powers docs for 20,000+ companies, reaching 100M+ people a year. We just raised a $45M Series B led by a16z and Salesforce Ventures to keep building the knowledge layer for the agent era.